/**
* MySQL supports a special form of UNION:
* (select ...)
* union
* (select ...)
*
* This function handles this query syntax. Only one such subquery
* is supported in each UNION block. (select)(select)union(select) is not legal.
* The extra queries will be silently ignored.
*/
protected function processMySQLUnion($queries)
{
$unionTypes = array('UNION', 'UNION ALL');
foreach ($unionTypes as $unionType) {
if (empty($queries[$unionType])) {
continue;
}
foreach ($queries[$unionType] as $key => $tokenList) {
foreach ($tokenList as $z => $token) {
$token = trim($token);
if ($token === "") {
continue;
}
// starts with "(select"
if (preg_match("/^\\(\\s*select\\s*/i", $token)) {
$processor = new DefaultProcessor();
$queries[$unionType][$key] = $processor->process($this->removeParenthesisFromStart($token));
break;
}
$processor = new SQLProcessor();
$queries[$unionType][$key] = $processor->process($queries[$unionType][$key]);
break;
}
}
}
// it can be parsed or not
return $queries;
}
/**
* It parses the given SQL statement and generates a detailled
* output array for every part of the statement. The method can
* also generate [position] fields within the output, which hold
* the character position for every statement part. The calculation
* of the positions needs some time, if you don't need positions in
* your application, set the parameter to false.
*
* @param String $sql The SQL statement.
* @param boolean $calcPositions True, if the output should contain [position], false otherwise.
*
* @return array An associative array with all meta information about the SQL statement.
*/
public function parse($sql, $calcPositions = false)
{
$processor = new DefaultProcessor();
$queries = $processor->process($sql);
// calc the positions of some important tokens
if ($calcPositions) {
$calculator = new PositionCalculator();
$queries = $calculator->setPositionsWithinSQL($sql, $queries);
}
// store the parsed queries
$this->parsed = $queries;
return $this->parsed;
}
protected function processFromExpression(&$parseInfo)
{
$res = array();
// exchange the join types (join_type is save now, saved_join_type holds the next one)
$parseInfo['join_type'] = $parseInfo['saved_join_type'];
// initialized with JOIN
$parseInfo['saved_join_type'] = $parseInfo['next_join_type'] ? $parseInfo['next_join_type'] : 'JOIN';
// we have a reg_expr, so we have to parse it
if ($parseInfo['ref_expr'] !== false) {
$unparsed = $this->splitSQLIntoTokens($this->removeParenthesisFromStart($parseInfo['ref_expr']));
// here we can get a comma separated list
foreach ($unparsed as $k => $v) {
if ($this->isCommaToken($v)) {
$unparsed[$k] = "";
}
}
$processor = new ExpressionListProcessor();
$parseInfo['ref_expr'] = $processor->process($unparsed);
}
// there is an expression, we have to parse it
if (substr(trim($parseInfo['table']), 0, 1) == '(') {
$parseInfo['expression'] = $this->removeParenthesisFromStart($parseInfo['table']);
if (preg_match("/^\\s*select/i", $parseInfo['expression'])) {
$processor = new DefaultProcessor();
$parseInfo['sub_tree'] = $processor->process($parseInfo['expression']);
$res['expr_type'] = ExpressionType::SUBQUERY;
} else {
$tmp = $this->splitSQLIntoTokens($parseInfo['expression']);
$parseInfo['sub_tree'] = $this->process($tmp);
$res['expr_type'] = ExpressionType::TABLE_EXPRESSION;
}
} else {
$res['expr_type'] = ExpressionType::TABLE;
$res['table'] = $parseInfo['table'];
$res['no_quotes'] = $this->revokeQuotation($parseInfo['table']);
}
$res['alias'] = $parseInfo['alias'];
$res['join_type'] = $parseInfo['join_type'];
$res['ref_type'] = $parseInfo['ref_type'];
$res['ref_clause'] = $parseInfo['ref_expr'];
$res['base_expr'] = trim($parseInfo['expression']);
$res['sub_tree'] = $parseInfo['sub_tree'];
return $res;
}
protected function processDefault($token)
{
$processor = new DefaultProcessor($this->options);
return $processor->process($token);
}
protected function processTopLevel($sql)
{
$processor = new DefaultProcessor();
return $processor->process($sql);
}
private function revokeQuotation($token)
{
$defProc = new DefaultProcessor();
return $defProc->revokeQuotation($token);
}
protected function processSQLDefault($unparsed)
{
$processor = new DefaultProcessor();
return $processor->process($unparsed);
}
protected function processColumns($cols)
{
if ($cols === false) {
return $cols;
}
if ($cols[0] === '(' && substr($cols, -1) === ')') {
$parsed = array('expr_type' => ExpressionType::BRACKET_EXPRESSION, 'base_expr' => $cols, 'sub_tree' => false);
}
$cols = $this->removeParenthesisFromStart($cols);
if (stripos($cols, 'SELECT') === 0) {
$processor = new DefaultProcessor();
$parsed['sub_tree'] = array(array('expr_type' => ExpressionType::QUERY, 'base_expr' => $cols, 'sub_tree' => $processor->process($cols)));
} else {
$processor = new ColumnListProcessor();
$parsed['sub_tree'] = $processor->process($cols);
$parsed['expr_type'] = ExpressionType::COLUMN_LIST;
}
return $parsed;
}
public function process($tokens)
{
$resultList = array();
$skip_next = false;
$prev = new ExpressionToken();
foreach ($tokens as $k => $v) {
$curr = new ExpressionToken($k, $v);
if ($curr->isWhitespaceToken()) {
continue;
}
if ($skip_next) {
// skip the next non-whitespace token
$skip_next = false;
continue;
}
/* is it a subquery? */
if ($curr->isSubQueryToken()) {
$processor = new DefaultProcessor();
$curr->setSubTree($processor->process($this->removeParenthesisFromStart($curr->getTrim())));
$curr->setTokenType(ExpressionType::SUBQUERY);
} elseif ($curr->isEnclosedWithinParenthesis()) {
/* is it an in-list? */
$localTokenList = $this->splitSQLIntoTokens($this->removeParenthesisFromStart($curr->getTrim()));
if ($prev->getUpper() === 'IN') {
foreach ($localTokenList as $k => $v) {
$tmpToken = new ExpressionToken($k, $v);
if ($tmpToken->isCommaToken()) {
unset($localTokenList[$k]);
}
}
$localTokenList = array_values($localTokenList);
$curr->setSubTree($this->process($localTokenList));
$curr->setTokenType(ExpressionType::IN_LIST);
} elseif ($prev->getUpper() === 'AGAINST') {
$match_mode = false;
foreach ($localTokenList as $k => $v) {
$tmpToken = new ExpressionToken($k, $v);
switch ($tmpToken->getUpper()) {
case 'WITH':
$match_mode = 'WITH QUERY EXPANSION';
break;
case 'IN':
$match_mode = 'IN BOOLEAN MODE';
break;
default:
}
if ($match_mode !== false) {
unset($localTokenList[$k]);
}
}
$tmpToken = $this->process($localTokenList);
if ($match_mode !== false) {
$match_mode = new ExpressionToken(0, $match_mode);
$match_mode->setTokenType(ExpressionType::MATCH_MODE);
$tmpToken[] = $match_mode->toArray();
}
$curr->setSubTree($tmpToken);
$curr->setTokenType(ExpressionType::MATCH_ARGUMENTS);
$prev->setTokenType(ExpressionType::SIMPLE_FUNCTION);
} elseif ($prev->isColumnReference() || $prev->isFunction() || $prev->isAggregateFunction()) {
// if we have a colref followed by a parenthesis pair,
// it isn't a colref, it is a user-function
// TODO: this should be a method, because we need the same code
// below for unspecified tokens (expressions).
$localExpr = new ExpressionToken();
$tmpExprList = array();
foreach ($localTokenList as $k => $v) {
$tmpToken = new ExpressionToken($k, $v);
if (!$tmpToken->isCommaToken()) {
$localExpr->addToken($v);
$tmpExprList[] = $v;
} else {
// an expression could have multiple parts split by operands
// if we have a comma, it is a split-point for expressions
$tmpExprList = array_values($tmpExprList);
$localExprList = $this->process($tmpExprList);
if (count($localExprList) > 1) {
$localExpr->setSubTree($localExprList);
$localExpr->setTokenType(ExpressionType::EXPRESSION);
$localExprList = $localExpr->toArray();
$localExprList['alias'] = false;
$localExprList = array($localExprList);
}
if (!$curr->getSubTree()) {
$curr->setSubTree($localExprList);
} else {
$tmpExprList = $curr->getSubTree();
$curr->setSubTree(array_merge($tmpExprList, $localExprList));
}
$tmpExprList = array();
$localExpr = new ExpressionToken();
}
}
$tmpExprList = array_values($tmpExprList);
$localExprList = $this->process($tmpExprList);
if (count($localExprList) > 1) {
$localExpr->setSubTree($localExprList);
$localExpr->setTokenType(ExpressionType::EXPRESSION);
$localExprList = $localExpr->toArray();
$localExprList['alias'] = false;
$localExprList = array($localExprList);
}
if (!$curr->getSubTree()) {
$curr->setSubTree($localExprList);
} else {
$tmpExprList = $curr->getSubTree();
$curr->setSubTree(array_merge($tmpExprList, $localExprList));
}
$prev->setSubTree($curr->getSubTree());
if ($prev->isColumnReference()) {
$prev->setTokenType(ExpressionType::SIMPLE_FUNCTION);
$prev->setNoQuotes(null);
}
array_pop($resultList);
$curr = $prev;
}
// we have parenthesis, but it seems to be an expression
if ($curr->isUnspecified()) {
// TODO: the localTokenList could contain commas and further expressions,
// we must handle that like function parameters (see above)!
// this should solve issue 51
$curr->setSubTree($this->process($localTokenList));
$curr->setTokenType(ExpressionType::BRACKET_EXPRESSION);
}
} elseif ($curr->isVariableToken()) {
# a variable
# it can be quoted
$curr->setTokenType($this->getVariableType($curr->getUpper()));
$curr->setSubTree(false);
$curr->setNoQuotes(trim(trim($curr->getToken()), '@'), "`'\"");
} else {
/* it is either an operator, a colref or a constant */
switch ($curr->getUpper()) {
case '*':
$curr->setSubTree(false);
// o subtree
// single or first element of expression list -> all-column-alias
if (empty($resultList)) {
$curr->setTokenType(ExpressionType::COLREF);
break;
}
// if the last token is colref, const or expression
// then * is an operator
// but if the previous colref ends with a dot, the * is the all-columns-alias
if (!$prev->isColumnReference() && !$prev->isConstant() && !$prev->isExpression() && !$prev->isBracketExpression() && !$prev->isAggregateFunction() && !$prev->isVariable()) {
$curr->setTokenType(ExpressionType::COLREF);
break;
}
if ($prev->isColumnReference() && $prev->endsWith(".")) {
$prev->addToken('*');
// tablealias dot *
continue 2;
// skip the current token
}
$curr->setTokenType(ExpressionType::OPERATOR);
break;
case ':=':
case 'AND':
case '&&':
case 'BETWEEN':
case 'AND':
case 'BINARY':
case '&':
case '~':
case '|':
case '^':
case 'DIV':
case '/':
case '<=>':
case '=':
case '>=':
case '>':
case 'IS':
case 'NOT':
case '<<':
case '<=':
case '<':
case 'LIKE':
case '%':
case '!=':
case '<>':
case 'REGEXP':
case '!':
case '||':
case 'OR':
case '>>':
case 'RLIKE':
case 'SOUNDS':
case 'XOR':
case 'IN':
$curr->setSubTree(false);
$curr->setTokenType(ExpressionType::OPERATOR);
break;
case 'NULL':
$curr->setSubTree(false);
$curr->setTokenType(ExpressionType::CONSTANT);
break;
case '-':
case '+':
// differ between preceding sign and operator
$curr->setSubTree(false);
if ($prev->isColumnReference() || $prev->isFunction() || $prev->isAggregateFunction() || $prev->isConstant() || $prev->isSubQuery() || $prev->isExpression() || $prev->isBracketExpression() || $prev->isVariable()) {
$curr->setTokenType(ExpressionType::OPERATOR);
} else {
$curr->setTokenType(ExpressionType::SIGN);
}
break;
default:
$curr->setSubTree(false);
switch ($curr->getToken(0)) {
case "'":
case '"':
// it is a string literal
$curr->setTokenType(ExpressionType::CONSTANT);
break;
case '`':
// it is an escaped colum name
$curr->setTokenType(ExpressionType::COLREF);
$curr->setNoQuotes($curr->getToken());
break;
default:
if (is_numeric($curr->getToken())) {
if ($prev->isSign()) {
$prev->addToken($curr->getToken());
// it is a negative numeric constant
$prev->setTokenType(ExpressionType::CONSTANT);
continue 3;
// skip current token
} else {
$curr->setTokenType(ExpressionType::CONSTANT);
}
} else {
$curr->setTokenType(ExpressionType::COLREF);
$curr->setNoQuotes($curr->getToken());
}
break;
}
}
}
/* is a reserved word? */
if (!$curr->isOperator() && !$curr->isInList() && !$curr->isFunction() && !$curr->isAggregateFunction() && PHPSQLParserConstants::isReserved($curr->getUpper())) {
if (PHPSQLParserConstants::isAggregateFunction($curr->getUpper())) {
$curr->setTokenType(ExpressionType::AGGREGATE_FUNCTION);
$curr->setNoQuotes(null);
} elseif ($curr->getUpper() === 'NULL') {
// it is a reserved word, but we would like to set it as constant
$curr->setTokenType(ExpressionType::CONSTANT);
} else {
if (PHPSQLParserConstants::isParameterizedFunction($curr->getUpper())) {
// issue 60: check functions with parameters
// -> colref (we check parameters later)
// -> if there is no parameter, we leave the colref
$curr->setTokenType(ExpressionType::COLREF);
} elseif (PHPSQLParserConstants::isFunction($curr->getUpper())) {
$curr->setTokenType(ExpressionType::SIMPLE_FUNCTION);
$curr->setNoQuotes(null);
} else {
$curr->setTokenType(ExpressionType::RESERVED);
$curr->setNoQuotes(null);
}
}
}
// issue 94, INTERVAL 1 MONTH
if ($curr->isConstant() && PHPSQLParserConstants::isParameterizedFunction($prev->getUpper())) {
$prev->setTokenType(ExpressionType::RESERVED);
$prev->setNoQuotes(null);
}
if ($prev->isConstant() && PHPSQLParserConstants::isParameterizedFunction($curr->getUpper())) {
$curr->setTokenType(ExpressionType::RESERVED);
$curr->setNoQuotes(null);
}
if ($curr->isUnspecified()) {
$curr->setTokenType(ExpressionType::EXPRESSION);
$curr->setNoQuotes(null);
$curr->setSubTree($this->process($this->splitSQLIntoTokens($curr->getTrim())));
}
$resultList[] = $curr;
$prev = $curr;
}
// end of for-loop
return $this->toArray($resultList);
}